I build data systems that survive real users, audits, and handover.

Sydney-based Senior Software Engineer and Data & Systems Specialist. Current work spans SQL Server, Power BI, Next.js, secure school data products, applied ML, and agentic AI workflows documented in public.

10+ yrs
shipping software, data, and automation systems
2,000 users
pentested parent portal, shipped solo in 8 weeks
200+
ReviewPulse commits turning coursework into an inspectable NLP lab
1,400+
commits on an open-source agentic study pipeline

Current proof

Recent proof across secure education systems, SQL/Power BI data work, applied ML, and agentic AI workflows documented in public.

Education data products

Secure systems for real school operations

Current work is production-facing: parent access, student-profile views, support workflows, academic reporting, and portal migration support in a regulated school environment.

  • Pentested 2,000-parent portal with authorization rechecked at the data boundary.
  • Student360-style profile surfaces built around mock/public safety and local-only live-data gates.
  • FreshService kiosk and Schoolbox data support shipped with operational handover in mind.
Read education case study

Applied ML systems

Models treated as products, not notebook trophies

Recent ML work focuses on leakage-safe evaluation, model selection, artifact provenance, deployment trade-offs, and interfaces that expose where predictions fail.

  • ReviewPulse v3: aspect-based sentiment lab with six model paths and token-level evidence.
  • Sommelier API: 22 model/treatment runs before replacing the shipped classifier.
  • Churn analysis: optimized for recall and cost trade-offs instead of vanity accuracy.
See ML proof

Agentic AI + security

Agentic AI workflows with public receipts

I use agents to compress throughput, but I document where the human owns judgment: assessment decisions, source truth, defensive controls, and privacy boundaries.

  • Public agentic study pipeline: map, notes, compression, active recall, one-pagers, assessment checks.
  • Technical reconstruction of the OpenAI-Hugging Face agent incident from public sources.
  • Preference for explicit contracts, redaction, auditability, and default-deny thinking.
Read agentic AI write-up

What I do

Secure education systems

School-facing products and internal tools where authorization, privacy, auditability, and handover are part of the design - not cleanup work after launch.

Next.jsTypeScriptSQL ServerMicrosoft Entra IDPython

Data engineering & analytics

SQL Server, Power BI, ETL, and reporting workflows built close to operations: finance, enrolments, academic data, support, and stakeholder dashboards.

SQL ServerT-SQLPower BIPower QueryETL

Applied ML systems

Models handled like production assets: evaluation contracts, artifact provenance, failure visibility, and interfaces that make trade-offs inspectable.

PyTorchscikit-learnTransformersStreamlitFastAPI

Agentic AI delivery

Claude Code and Codex workflows used with explicit source truth, review gates, redaction, and public write-ups showing what the agent did and what stayed human-owned.

Claude CodeCodexGitHubDev.toAI security

Core stack

Production systems

Next.jsTypeScriptPythonSQL ServerMicrosoft Entra ID

Data & analytics

T-SQLPower BIPower QuerySSIS / ETLpandasPySpark

Applied ML

PyTorchscikit-learnTransformersStreamlitFastAPI

Delivery & agents

DockerGitHub ActionsNginxSentryClaude Code / Codex

Timeline at a glance

Key career milestones and technical achievements

  1. 2023–2024

    Designed and launched Konquista, a Django + Celery/Redis WhatsApp automation platform powering 30K+ monthly messages across all clinics. Expanded Python expertise across FastAPI, Flask, and Streamlit for production-grade ML and automation pipelines.

  2. 2024

    Relocated to Sydney and continued delivering for international clients across time zones. Strengthened backend architecture, designed scalable Python systems, and completed Stanford’s Machine Learning Specialization.

  3. 2025

    Pursuing a Master’s in Software Engineering & AI while expanding full-stack capabilities. Shipped Wedstack (Next.js + GraphQL + Stripe), built AI tools on OpenAI, and published 30+ engineering write-ups on dev.to.

  4. 2026

    Joined St Catherine’s School, Sydney and moved into Data & Systems Specialist work: secure Next.js products, SQL Server pipelines, Power BI reporting, academic-data modernisation, FreshService and Schoolbox support, and Student360-style data surfaces - while completing ML, Deep Learning, and Big Data subjects in the Master’s.